Chat log from the meeting on 2007-11-06

[10:16] You: hey
[10:17] You: you guys are early
[10:17] You: hehe
[10:17] You: its neb
[10:17] Darok Kaminski: hi
[10:18] Teravus Ousley: back
[10:19] Teravus Ousley: :D
[10:19] Teravus Ousley: If you sit on a prim, you won't be able to talk. :D We found that out yesterday :D
[10:19] You: hehe yea
[10:24] You: hey
[10:24] danx2 danx0r: howdy
[10:24] Teravus Ousley: Hello Dan
[10:24] danx2 danx0r: looking sharp these days!
[10:25] Teravus Ousley: heh
[10:25] You: hit play on video
[10:25] You: if you can watch video that is
[10:25] You: hehe
[10:25] danx2 danx0r: wtf?
[10:25] danx2 danx0r: someone added video?
[10:25] You: i did
[10:25] Darok Kaminski: hello danx0r
[10:25] You: this is neb
[10:26] danx2 danx0r: hi
[10:26] danx2 danx0r: how do I play video?
[10:26] You: do you have quicktime?
[10:26] danx2 danx0r: not sure I have QT
[10:26] You: ah ok yea you need QT
[10:26] danx2 danx0r: how do I check?
[10:26] Teravus Ousley: I don't have QT on this computer.. I keep most things off of it.
[10:26] You: you can start by going into preferences
[10:26] You: on Audio/Video tab
[10:27] You: and make sure play video is checked
[10:27] danx2 danx0r: it is
[10:27] You: if quicktime is installed you should get video control at the bottom
[10:27] You: a play and stop button
[10:27] danx2 danx0r: nope, I'll try later
[10:27] danx2 danx0r: who put this in?
[10:27] You: i made a video of me and teravus kicking prims around
[10:27] danx2 danx0r: I used to do alot of video code
[10:27] You: its always worked in OpenSIM
[10:28] danx2 danx0r: I worked on the theora codec (ogg video)
[10:28] You: i made this video
[10:28] Chillken Proto: hehe, nice :)
[10:28] danx2 danx0r: heh but today I got a msg saying the place supports video. Guess that's cuz we have a video here
[10:28] Teravus Ousley: :D
[10:28] Darok Kaminski: The video is very cool :D
[10:30] You: hehe dan
[10:30] Teravus Ousley: If you sit on a prim, you won't be able to chat
[10:30] danx2 danx0r: once we have persistent shape+clothes, then the chicks will hang out here too
[10:31] danx2 danx0r: and all bets are off
[10:31] Neas Bade: yeh, what's up with +sit -chat?
[10:31] You: lol
[10:31] You: yea not sure
[10:31] danx2 danx0r: I've never sat - does it work?
[10:31] Neas Bade: it turns out it does route it to IRC, just not in world
[10:31] You: Is IRC relay fixed now?
[10:32] danx2 danx0r: that's sitting?
[10:32] Teravus Ousley: it's hard to say
[10:33] danx2 danx0r: there seems to be a conflict between sitting and physics
[10:33] Neas Bade: danx0r
[10:33] Neas Bade: yes
[10:33] Neas Bade: if you want to sit on something you need to make it phantom
[10:33] Teravus Ousley: heh, remember.. if you don't want to crash the sim.. don't resize a meshed prim that isn't a cube :D
[10:33] danx2 danx0r: hmmm
[10:33] danx2 danx0r: didn't gerhard fix that?
[10:33] Chillken Proto: the autopilot for sitting has some trouble
[10:33] danx2 danx0r: I saw a patch
[10:34] Teravus Ousley: this is 2290, 2291 (my patch) fixed that.
[10:34] Chillken Proto: it seems like in SL they move you into position manually if you're close to the object
[10:34] danx2 danx0r: ahh, ok
[10:34] danx2 danx0r: welcome btw
[10:34] Neas Bade: come on into the building folks :)
[10:34] Teravus Ousley: thx :D
[10:34] danx2 danx0r: pilates starting? I need it
[10:35] Neas Bade: heh
[10:35] You: rofl
[10:35] Neas Bade: nebadon can you make a physical 2m cube to kick around?
[10:35] You: it doesnt move as well
[10:35] danx2 danx0r: I have this nagging compulsion to start working on a client
[10:35] Teravus Ousley: hehe
[10:35] You: 2m is kinda big for kicking
[10:36] danx2 danx0r: oh is there a cube to kick around?
[10:36] Neas Bade: well, you can push it around
[10:36] danx2 danx0r: so if I'm this fat, I should impart more momentum to objects
[10:36] Teravus Ousley: lol, yes.. you have to shrink it some.. or it won't move much
[10:36] danx2 danx0r: where's the friction?
[10:37] danx2 danx0r: is this ODE or bulet?
[10:37] Neas Bade: it's there, it's just small
[10:37] Teravus Ousley: heh. It's a lot of mass
[10:37] You: its coming right for us
[10:37] Neas Bade: ODE is my guess
[10:37] danx2 danx0r: look at how the client interp works/doesn't
[10:37] You: this is running BulletX
[10:37] danx2 danx0r: teravus, you must have set fricton very low?
[10:37] You: ODE crashes this sim
[10:38] You: this is not ODE
[10:38] You: this is bulletx
[10:38] danx2 danx0r: ahh
[10:38] danx2 danx0r: hah!
[10:38] Neas Bade: ah, I was running ODE earlier
[10:38] danx2 danx0r: no wonder
[10:38] Neas Bade: and it was working quite well
[10:38] Neas Bade: same sort of friction characteristics
[10:38] Teravus Ousley: I didn't know you could jump in BulletX now :D
[10:38] danx2 danx0r: ODE was doing this?
[10:38] Teravus Ousley: neat.
[10:39] danx2 danx0r: I can't believe the cube would travel this far in ODE as presently configured
[10:39] Neas Bade: it did in my environment
[10:39] You: ODE doesnt slide smooth at all
[10:39] You: it more jumps
[10:41] danx2 danx0r: I want to see if you stand on it, and push it, will you stay on top?
[10:41] danx2 danx0r: heh you gotta walk to keep up
[10:41] Darok Kaminski: This should be ODE not BulletX cause in BulletX rotations are not working in this moment
[10:41] Neas Bade: looks lik no
[10:42] danx2 danx0r: there's no friction at all in bullet AFAICS
[10:42] Teravus Ousley: I was going to say..
[10:42] danx2 danx0r: I wonder why prim motion is smoother than ODE
[10:42] Teravus Ousley: are you *sure* this isn't ODE?
[10:43] You: let me check
[10:43] Teravus Ousley: it acts a lot like ODE
[10:43] Neas Bade: this looks like ODE did for me earlier today
[10:43] Darok Kaminski: I was looking to change values of friction of objects in BulletX, increasing it, but it has not effect
[10:43] You: last i heard it was bullet
[10:43] Darok Kaminski: It should be ODE
[10:43] danx2 danx0r: darok, there mus tbe something wrong then
[10:43] danx2 danx0r: no this is bullet
[10:43] Teravus Ousley: cfk said he was going to enable ODE for the meeting unless it crashes
[10:43] You: actully
[10:43] You: it appears to be running ODE
[10:44] You: ok
[10:44] You: last night he said he changed it to Bullet
[10:44] danx2 danx0r: look at that
[10:44] You: i guess he changed it back
[10:44] Teravus Ousley: Just massively heavy objects
[10:44] Neas Bade: yep
[10:44] Neas Bade: it's pretty awesomely functional vs. any previous state
[10:44] danx2 danx0r: ahh I know what's happening
[10:45] danx2 danx0r: it's to do with the 'friction approximation model'
[10:45] danx2 danx0r: we must be using the simple one that has a fixed restoring force regardless of mass or pressure
[10:45] danx2 danx0r: ahh, and it's complicated by client update issues
[10:45] Neas Bade: honestly, the sliding is kind of fun
[10:45] danx2 danx0r: you all seeing it same, right? kick it, it moves, snaps back
[10:46] Teravus Ousley: see.. friction on smaller objects works.. I think the big one is just 'really' heavy
[10:46] danx2 danx0r: so in ODE it just moves a bit
[10:46] danx2 danx0r: teravus -- the friction model you set in the collision code
[10:47] Teravus Ousley: heh
[10:47] danx2 danx0r: if it's ODE, I thought standing on the prim should work
[10:47] Teravus Ousley: normally there's also rotational velocity.. which isn't geting sent to the client
[10:47] danx2 danx0r: d'oh no --
[10:47] Teravus Ousley: heh
[10:48] Neas Bade: not if there isn't enough friction, right?
[10:48] danx2 danx0r: I have that stupid hack to keep me in place if no cursor keys are pushed
[10:48] danx2 danx0r: friction is irrelevant here -- I force the avatar not to move
[10:48] Neas Bade: ah ok
[10:48] danx2 danx0r: remember it used to slip around like bullet?
[10:49] Teravus Ousley: someone took out dan
[10:49] Teravus Ousley: someone took out dan.. heh
[10:49] Darok Kaminski: ei, Could we play soccer? :D
[10:50] Teravus Ousley: There's one more bug that this revision has.. Geoms are not getting disposed when resizing a prim
[10:51] Teravus Ousley: so, you'll notice if you resize a prim and set it physical.. it'll collide immdiately with the 'ghost prim' geom.
[10:51] Neas Bade: it would be cool if the engine let things get over this lip
[10:51] Teravus Ousley: well.. it's a cube
[10:51] Neas Bade: dan, teravus, any idea if that is hard or easy?
[10:52] Teravus Ousley: as far as the physics engine is concerned.. it's a cube..
[10:52] Neas Bade: right, but a real cube would tip
[10:52] Teravus Ousley: not one that heavy :D
[10:52] Teravus Ousley: try a smaller one
[10:52] Neas Bade: as center of mass is higher
[10:54] Teravus Ousley: hehe
[10:54] You: sim is holding up good
[10:55] Teravus Ousley: the way the code works.. is it tries to give prim half as priority on physics interaction
[10:55] Teravus Ousley: .. avatar movement first.. prim movement second
[10:56] You: lol
[10:56] Teravus Ousley: It's got a time in milliseconds to drop frames gracefully.. and the 'ignore prim collissions routine' is half of that.
[10:57] Neas Bade: yeh, other than a lot of console messages about not storing physical object in db, I was doing pretty good
[10:57] Neas Bade: plaza, what's the cpu usage?
[10:58] Teravus Ousley: yeah. Can't store the prim in the database once they become physical.. otherwise it lags in SQLite commits.. because it 'backs the prim up' on change
[10:58] Teravus Ousley: physical prim change a *LOT*
[10:59] Neas Bade: teravus, definitely understood. I'll look at something smarter in that
[10:59] danx2 danx0r: boss just started an impromptu meeting - later folks
[10:59] Teravus Ousley: ok.. take care
[10:59] Neas Bade: it would be nice to not put soo many messages on the console
[11:00] Neas Bade: as I get about 20 / 3 seconds
[11:00] You: hehe i put the video screen on a phyiscal prim
[11:00] Teravus Ousley: hehe.. imagine if SQLite was storing that :P
[11:00] You: hit play on video
[11:00] Neas Bade: Teravus, definitely :)
[11:00] You: thats trippy
[11:01] Teravus Ousley: we need a better way to send terse updates from physics anyway
[11:01] Teravus Ousley: .. the 'blip' .. 'blip'.. ... 'blip'... update system isn't cutting it :D
[11:01] Darok Kaminski: Lunch awaits me...
[11:01] Neas Bade: hoping we can get MW in here and actually talk about your proposal there
[11:01] Darok Kaminski: so i have to go :P
[11:02] Teravus Ousley: ok, take care Darok :D
[11:02] Darok Kaminski: thx :D
[11:02] Neas Bade: take care
[11:02] Darok Kaminski: see you!
[11:02] Neas Bade: hey MW
[11:02] Teravus Ousley: Hello :D
[11:02] More Wright: hi
[11:04] More Wright: hmm can't push him off his stool
[11:04] Teravus Ousley: Is he a linked object?
[11:04] Neas Bade: or physical?
[11:04] Neas Bade: not physical
[11:05] Teravus Ousley: technically.. linked objects are not supported at the moment
[11:05] You: hehe like my hippo
[11:05] Neas Bade: actually, not even linked it would seem
[11:05] Teravus Ousley: heh
[11:05] You: did you move the logo ?
[11:05] You: MW?
[11:05] Neas Bade: I made it physical, and it fell off
[11:06] More Wright: yeah, I didn't touch it
[11:06] More Wright: honestly
[11:06] Neas Bade: stupid gravity, my mortal enemy
[11:06] Neas Bade: it was my bad
[11:06] Teravus Ousley: hehe
[11:06] Teravus Ousley: be careful about cutting prim
[11:06] Teravus Ousley: .. don't resize it.. or it'll crash the sim on this revision
[11:06] Adam Frisby: Allo allo
[11:06] Neas Bade: there is also one of those phantom spaces here
[11:07] Teravus Ousley: Resizing cut 'non' cubes crashes
[11:07] Neas Bade: hey Adam
[11:07] Teravus Ousley: Yeah.. gotta update the sim to the one with the patch :D
[11:07] You: yea i noticed that yesterday
[11:07] You: i was resizing cut cylinder
[11:07] You: sim died
[11:08] Teravus Ousley: next revision after this.. fixes that.. and the 'invisible' prim
[11:09] Neas Bade: well, while we have Teravus and MW and others here, I suggest a lively discussion on physics is appropriate
[11:09] Neas Bade: especially seeing how much progress has been made
[11:09] Teravus Ousley would dance.. but doesn't have any dance animations
[11:11] danx2 danx0r: this meeting is boring
[11:11] You: lol
[11:11] danx2 danx0r: industrial design
[11:11] You: oh your robot meeting
[11:11] Neas Bade: :)
[11:11] Teravus Ousley sets some prim 'non' physical.
[11:11] danx2 danx0r: some dweeb selling us on his SLS skillz
[11:12] Neas Bade: Teravus, want to give us a brief overview of your approach #1 that MW was positive on?
[11:12] danx2 danx0r: robotics is on a slower track than VW's
[11:12] You: hehe man my avatar is disturbing
[11:13] Teravus Ousley: Well.. I was opening up discussion on revamping the physics system.. and I outlined a few of the requirements that I knew about
[11:13] Teravus Ousley: The one that MW liked was 'events' based.. where the physics plugin would raise events .. that caused the scene.. to do something
[11:14] More Wright: brb
[11:14] Teravus Ousley: events based is a physics oriented approach
[11:14] Teravus Ousley: our current approach is a 'scene oriented' approach .. where the scene can 'peek' into the physics scene when it wants to or needs to
[11:15] Teravus Ousley: .. The events approach is what MV liked though.. so.. one challenge with the events approach.. is the physics system generates a *lot* of events
[11:16] Neas Bade: right, how does one handle throttling in such an environment
[11:16] Teravus Ousley: we'll need to tone that down some or risk using up a *ton* of processor time on events.
[11:16] Teravus Ousley: one suggestion that was made was using a 'subscribe to events' model.. where one object could subscribe to the events of another.. by type.. and by object.
[11:17] Neas Bade: how often would events be emitted?
[11:17] Teravus Ousley: that's the tricky part.. you'd have to tweak it for a good balance between simulation and processor time.
[11:17] Adam Frisby: What about emitting on demand?
[11:17] Adam Frisby: IE: Physics.EmitEvents()
[11:17] Adam Frisby: we can call that once per scene frame.
[11:18] Neas Bade: aren't you just back into a poll model then?
[11:18] Adam Frisby: Not quite.
[11:18] Adam Frisby: Still bind to the events individually.
[11:18] Adam Frisby: and it is pushed out, but the pushing is done in sync with the rest of the world.
[11:19] Teravus Ousley listens to comments.
[11:19] Teravus Ousley: that's about as far as I got on it so far..
[11:19] Teravus Ousley: It's open for discussion.
[11:20] More Wright: yeah I think the talk before was of Scene doing a call into physics engine each frame, then on that action, the physics engine firing off its events for that frame, not sure how well we could get that to work
[11:20] danx2 danx0r: I like that model
[11:20] Adam Frisby: Can we make a eventqueue<> ?
[11:21] More Wright: but if I remember that was the general idea, at the time of these talks, it was really just me, lbsa71 , Adam, and Dan involved
[11:21] Adam Frisby: where the events fire into the queue, but the queue doesnt run until we ask it to?
[11:22] danx2 danx0r: however, what is actually triggering these events -- the "physical" part of an object, or the object itself?
[11:22] Adam Frisby: physical part
[11:22] danx2 danx0r: not sure I like that part
[11:22] danx2 danx0r: I'm for a tighter coupling between the definition of objects and their physical behavior
[11:24] Teravus Ousley: k, we'll need to come up with a design scheme that physics can be retrofitted into
[11:24] danx2 danx0r: what I mean is, the way it's being described, the event model assumes that physics basically just tells objects what to do
[11:24] Teravus Ousley: as you can see.. right now it works..
[11:25] Teravus Ousley: just not as well as it could
[11:25] Teravus Ousley shuts up and lets dan speak
[11:25] danx2 danx0r: not sure I can explain my misgivings easily
[11:26] danx2 danx0r: I'm just a bit concerned that there is an artificial division into physics and non-physics stuff
[11:26] danx2 danx0r: question: how does scripting work? Is that an event model?
[11:26] Adam Frisby: Well, it's mostly for abstractions sake - if we could, I'd love to move physics into a regionmodule
[11:27] Adam Frisby: as far as I know, yes. Tedd is on IRC now I think
[11:27] danx2 danx0r: there is some similiarity between physics and scripts
[11:27] danx2 danx0r: they both want to control objects
[11:28] Neas Bade: though the time resolution isn't quite the same
[11:28] Teravus Ousley: scripting is an active process.. with 'threads' that process for multiple scripts on a LLSim. It generates events.. that get processed in an event queue.
[11:28] danx2 danx0r: teravus: physics is very similar, if you think about it
[11:28] danx2 danx0r: we think of it as monolithic
[11:29] Neas Bade: I guess a question is what happens when we peg out the environment in an event model
[11:29] danx2 danx0r: but in reality, it's localized -- physical objects only affect/are affected by certain other objects, which are 'near'
[11:29] Neas Bade: if you are doing queues, you have to be able to drop things from the queue
[11:30] danx2 danx0r: maybe the event model makes sense. I have to think about it
[11:30] More Wright: err I'm for keeping physics and non-physical de-coupled as much as possible
[11:30] Teravus Ousley: the LLScripting model.. doesn't drop scripting events... it just runs scripts at like 1/100th of their speed non pegged
[11:31] danx2 danx0r: what's pegged mean, sorry
[11:31] Neas Bade: cpu 10)%
[11:31] Neas Bade: 100%
[11:31] Teravus Ousley: massive amount of events hogging the CPU
[11:31] danx2 danx0r: ahh
[11:31] danx2 danx0r: LL slows down physical clock I believe
[11:31] danx2 danx0r: which makes sense
[11:31] Neas Bade: right, but if you are fully an event model, how does that work
[11:32] Teravus Ousley: I'm doing the same here now..
[11:32] danx2 danx0r: neas: you just slow down until events are handled
[11:32] Neas Bade: not saying it doesn't work, just asking
[11:32] danx2 danx0r: in theory, you could stop the clock, and just let all the events flush through
[11:32] Teravus Ousley: if the time between last frame is a lot.. assume that the system is busy with something and slow down the physics scene to accomodate
[11:32] danx2 danx0r: it won't be a fun user experience obviously
[11:32] danx2 danx0r: but if you drop events, you might get some wierd inconsistencies
[11:33] Charles Krinkeb: speaking of pegging. The CPU running this region is currently at 45% and we are consuming 300Mbytes of RAM.
[11:33] Teravus Ousley: one such weird consistency .. is the 'bounce back'
[11:33] Adam Frisby: Uhh, that's not a good sign charles.
[11:33] danx2 danx0r: though it should be architected so missed events can never cause a major problem
[11:33] Adam Frisby: Hrrm.
[11:33] Adam Frisby: Can we run this sim in a profiler?
[11:34] Neas Bade: what is this running on?
[11:34] Adam Frisby: I'd be very keen to work out where it's dying.
[11:34] Optik Philtr: that seems high
[11:34] Charles Krinkeb: Last week, the CPU was 2% with ODE off and running basicphysics.
[11:34] Adam Frisby: I'm getting bandwidth spikes here
[11:34] danx2 danx0r: blame it on ODE!
[11:34] Adam Frisby: IE 2-30kbps then suddenly 800.
[11:34] Charles Krinkeb: not blaming, just observing
[11:34] Adam Frisby: Yeah, I'd like to find out why it's doing it.
[11:35] Teravus Ousley: I don't see any physicsl objects besides ones that hit the edge and are now disabled
[11:35] Optik Philtr: I do not think its ODE I had 10 people on mine test grid at home and it barely went over 8%
[11:35] Neas Bade: I was getting 10 - 15% utilization on one core with 6 avatars playing pushing things around earlier in my test env
[11:35] Optik Philtr: something else may be up
[11:35] Adam Frisby: Hrrm.
[11:35] Optik Philtr: is it trying to re restore all our inventories repeatedly?
[11:35] Adam Frisby: If I can take a stab, I'm betting network might be at the heart of it.
[11:35] Neas Bade: also, there are at least 4 bouncing prims on the edge of reality
[11:35] Neas Bade: you can see them on the mini map
[11:36] Optik Philtr: yes i do
[11:36] Teravus Ousley: those prim are disabled in the physics engine.. their velocity never got set to zero though..
[11:36] Adam Frisby: Since this is a OpenSim office hour - maybe we should talk about finally implementing the NetworkManager to handle throttling / streamlining of updates.
[11:36] Teravus Ousley: and they are generating terse updates
[11:36] Neas Bade: Adam, go for it
[11:36] Teravus Ousley: that could be the bandwidth spike you are mentioning.. .. *shusshes*
[11:38] Adam Frisby: Well, a case is with say terrain
[11:38] Adam Frisby: If someone terraforms - that will send a lot of agentupdates at the user, of which some will not be needed (ie two rapid updates happen, we should only send the second one)
[11:38] Adam Frisby: Likewise, texture downloads arnt throttled so you get the whole texture as fast-as-possible.
[11:38] Teravus Ousley notes that physical prim should be updated when their position or rotation changes... instead right now the system is sending constant updates every 2 or 3 seconds.
[11:38] Adam Frisby: Packet resending is a little flawed too -- the idea would be to route outbound packets through a network manager class
[11:39] Adam Frisby: which we could send as much as we like at it, and it would only send the last set of authoritive updates and handle retransmission.
[11:39] Neas Bade: Teravus, that would explain all the console messages on that time frame
[11:39] Teravus Ousley notes that the terse update system and the backup system are tied together
[11:40] Neas Bade: right, I think I'll dig into the sqlite stuff to at least make it a little better in that case
[11:40] Finrod Meriman: how much of the traffic is updates vs textures?
[11:40] Adam Frisby: I dont think we can accurately tell right now
[11:41] Neas Bade: I think a network manager would provide a nice point to do a lot of such things
[11:41] Neas Bade: ug my gramar falling appart
[11:41] Neas Bade: sorry, it would also provide us with centralized statitistics for such things
[11:41] Teravus Ousley: actually.. there's one built into the client
[11:42] Teravus Ousley: .. open up the debug menus.. Control + D
[11:42] Neas Bade: right, but we need to actually populate some of the server stats :)
[11:42] Teravus Ousley: then go to Client ---> 'consoles'.. and click 'Frame'
[11:42] Finrod Meriman: are there any plugins for ethereal?
[11:43] Teravus Ousley: There's a LOT of great tools in the Console section
[11:44] More Wright: I agree we need a lot better packet management, I just worry its still a bit premature. There is a lot of things that still need implementing that will have a effect on how updates work
[11:44] Adam Frisby: I can see it becoming more difficult to integrate in later though.
[11:44] Adam Frisby: It's already going to be a bundle of fun.
[11:44] More Wright: we need a option in .ini to set day cycle length or turn night off
[11:44] Adam Frisby: Maybe we should look at moving all the SendXYZ() functions into the network manager?
[11:45] Neas Bade: MW that would be great
[11:45] Neas Bade: where is the day cycle code actually
[11:45] Neas Bade: I could dig in there
[11:45] Adam Frisby: It's in Update()
[11:45] Adam Frisby: I spotted it a while ago when I was refactoring it.
[11:45] Neas Bade: Day might make an interesting region module actually
[11:45] Adam Frisby: Why not
[11:45] Neas Bade: woah
[11:45] You: woops
[11:46] Adam Frisby: is everyone bouncing?
[11:46] Adam Frisby: Hahah
[11:46] More Wright: well personally I'm most likely in favour of the Scenepresence class (or a class inside that ) doing packet management for that avatar, the scenepresence should act as a avatars eyes and that includes deciding what to update the client about
[11:46] Teravus Ousley: heh,
[11:46] You: and it moved the prim
[11:46] Neas Bade: what was that??
[11:46] Adam Frisby: Alien ray beam
[11:46] You: my bad
[11:46] You: i clicked the ground
[11:46] You: rofl
[11:46] Neas Bade: take one step and you'll fall to the ground
[11:46] Adam Frisby: Dun dun dun
[11:46] Optik Philtr: oh noes! quicksand
[11:47] Adam Frisby: We so should obey the "material" prim properties and make bouncy prims. :D
[11:47] Adam Frisby: IE: rubber makes your prim bounce uncontrollably. :D
[11:47] Teravus Ousley: superball prim? :P
[11:47] Teravus Ousley: I've got a script on the LLGrid that does that :D
[11:48] Adam Frisby: Negative friction. :D
[11:48] Adam Frisby: That could be fun
[11:48] Teravus Ousley: keys off ground collisions and prim collisions.. and applies the opposite velocity impulse..
[11:49] Teravus Ousley: minus, of course some energy that you set to loose.. or make it gain energy on each bounce :D
[11:49] Charles Krinkeb: do we have any conclusions or action items for ode physics based on the carefully controlled tests occurring?
[11:49] Teravus Ousley: are you standing on MW's head?
[11:49] Teravus Ousley: heheh
[11:49] You: that might have been the 1st official OpenSIM head stand
[11:49] Teravus Ousley: good work, Adam :D
[11:49] Charles Krinkeb: Said differently, are there other things we need to talk about before we run out of time today?
[11:50] Finrod Meriman: adam: you sent out a bunch of notes re: v0.5
[11:50] Finrod Meriman: are we set on that?
[11:50] Adam Frisby: Uhh
[11:50] Adam Frisby: Which notes where? I deny everything!
[11:50] Neas Bade: I didn't see any notes
[11:50] Finrod Meriman: questions on the mailing list
[11:50] Adam Frisby: The grid ones?
[11:51] Finrod Meriman: a couple days ago
[11:51] Finrod Meriman: yeah
[11:51] Adam Frisby: Oh yeah - we need to pick that thread back up
[11:51] Adam Frisby: I'd like to nail down at least the basics:
[11:51] Adam Frisby: * What rough communication protocol are we standardising on?
[11:51] Adam Frisby: * Does everyone agree with the rough principles put in there?
[11:51] Adam Frisby: * What can we start implementing?
[11:52] Neas Bade: the grid asset stuff the tleiades did so far is pretty promissing as a prototype for doing more of these services as REST calls
[11:52] Teravus Ousley: I think we'll need more discussion regarding the mechansim of doing it.. but I agree with the principles
[11:53] Charles Krinkeb: On the grid asset stuff, are there mono / nant issues in Linux that need to be addressed?
[11:53] Neas Bade: all fixed by MW
[11:53] More Wright: yeah with network comms, I think we can in general split them into two/three or four sub parts. One Assets and inventory. I think we have a good idea of how they will work
[11:53] You: what about the map issue
[11:53] You: in linux servers on grid
[11:53] Adam Frisby: We need to draft a few overarching standards I think - like how we handle authentication there.
[11:53] You: any clues?
[11:54] Adam Frisby: I liked the suggestion of using HTTP Authentication combined with TLS/SSL
[11:54] Neas Bade: yes, auth needs to be worked out
[11:54] Adam Frisby: we just need to fix up our CertificatePolicy when handling HTTPS queries.
[11:54] Neas Bade: yep
[11:54] More Wright: 2: comms between "grid server" "user server" and regions, although I think we need to get past the idea of X server as just think in terms of services
[11:54] Finrod Meriman: how does openid fit in?
[11:54] Finrod Meriman: or does it
[11:54] Adam Frisby: It doesnt
[11:54] Adam Frisby: OpenID always struck me as a solution in need of a problem.
[11:54] More Wright: 3: inter region comms (which might be able to be grouped with grid comms)
[11:54] More Wright: and 4: IM's
[11:55] Adam Frisby: Let's generalise that a little:
[11:55] Adam Frisby: 1. Region<->Grid Domain
[11:55] Finrod Meriman: why is IM part of the architecture?
[11:55] Adam Frisby: 2. Region<->Region
[11:55] Adam Frisby: 3. Client<->Client
[11:55] Adam Frisby: 4. Region<->Client
[11:56] Neas Bade: for completeness, is there any Client<->Grid?
[11:56] Adam Frisby: I'd like to use mostly the same overarching protocol for all four of those, then have the ability to negotiate more compact connections for higher speed problems.
[11:56] Neas Bade: presumably login is still Client<->Grid
[11:56] Adam Frisby: I dont think so, except maybe to setup Client<->Client transactions.
[11:56] Adam Frisby: Hrrm, yeah login might be a case there.
[11:56] Adam Frisby: I do think the LL AWG plan has got most of this right in general principles, just not in practical ones.
[11:57] Adam Frisby: IE: EventQueueGet which I keep raising.
[11:57] Neas Bade: I do think that that categorization makes a lot of sense though
[11:57] More Wright: well CAPS should be client To grid, I don't think we should have all Caps handlers on the region servers
[11:57] Neas Bade: MW is there a good place to read up more on CAPS?
[11:57] Adam Frisby: Frankly I do wonder if we should really insist on following LL's plans there so completely.
[11:58] Adam Frisby: I'm not convinced they have most of this right when it comes to actual implementation.
[11:58] More Wright: Neas, most likely the best place at the moment is libsl's wiki and the logs of old Zero's office hours, at the weekend I'll try to write something up and get it on the mailing list
[11:59] Neas Bade: my personal opinion is to only diverge when there is a concrete rationale to. A lot of users and developers are part of the opensim effort because of the compatibility it provides
[11:59] More Wright: Adam, I'm all for not following LL plans when they don't make sense, but think its something we need to think more about, getting somethings off the region servers does make sense in some ways
[11:59] Charles Krinkeb: Are there any practical things we need to decide on in this forum before we finish for the day?
[11:59] You: is the LibSL restructuring complete now?
[12:00] Neas Bade: yeh, the only things that really seem to have come from Zero's office hours is moving more features to be client <-> grid, and doing more interfaces with REST
[12:00] More Wright: well I'm for keeping Client compatibility but that in no ways means internally we have to do things anything like LL do
[12:00] Neas Bade: both seem pretty sensible
[12:01] More Wright: and As long as we all remember that OpenSim isn't a SL clone
[12:01] Neas Bade: yep, agreed
[12:01] Adam Frisby: Question: is IM's working?
[12:01] Neas Bade: compatibility doesn't mean clone
[12:01] Adam Frisby: *tests*
[12:01] You: i dont think so adam
[12:02] More Wright: it did before your module, so if its not then its you who broke it
[12:02] Adam Frisby: Woot
[12:02] Adam Frisby: It's working.
[12:02] You: yea
[12:02] Teravus Ousley: someone send me an IM! :D
[12:02] Teravus Ousley: lol
[12:03] You: is IRC relay working?
[12:03] Teravus Ousley: everyone sent me an IM.. and someone tried to friend me.. but it's saying that I offereed friendship to myself
[12:03] Neas Bade: that was me
[12:03] Optik Philtr: Tera, live use to do that also
[12:03] Neas Bade: plaza, my irc relay works, I haven't seen if it is working here
[12:03] Adam Frisby: Hey at least it's not crashing when you offer friendship now. hehe.
[12:03] You: about a week and a few days ago IRC stopped working
[12:03] Charles Krinkeb: Do we have any thoughts on making OSgrid a bit more reliable and then making updates be a bit more infrequent and more controlled?
[12:03] You: I will try again
[12:03] You: hang on
[12:03] You: let me restart my server
[12:04] You: it seemed possibly to be a problem with connecting to EFNET
[12:04] Optik Philtr: yea sometimes there DNS handoff is uber slow
[12:04] Optik Philtr: and times out
[12:04] Teravus Ousley: hehe .. that bounce.. was the physics system dropping frames to accomodate something :D
[12:05] Charles Krinkeb: Teravus: As long as Wright Plaza doesnt crash more then once a day, I will keep running ODE here.
[12:05] Teravus Ousley: well.. I suggest that you get the latest revision.. it's got two important fixes.. one is a sim crasher.
[12:05] Optik Philtr: oops
[12:05] More Wright: Charles to be honest, I think once we start the big work on the new network protocols , osgrid and grids in general might get more unreliable before they get better
[12:06] You: Yea IRC is not working
[12:06] You: neas can you test IRC with EFNET
[12:06] Charles Krinkeb: Can we attempt to fix a grid server version that is reliable and I will keep osgrid at that version until you and the others are ready to move it forward?
[12:06] You: it just does not work at all
[12:06] Adam Frisby: I'm going to be putting some attention on the grid server soonish.
[12:07] Adam Frisby: most of you know I'm fiddling around with getting deepgrid running properly again (in theory it should be working now, but it needs maintainence)
[12:07] You: problem i see with that charles is that if we fix it at a certain build
[12:07] Neas Bade: honestly, I'd suggest a push forward on grid services instead of trying to fix up what we have
[12:07] You: we dont get good real world testing
[12:07] Adam Frisby: I'd like to push forward on replacing the grid services with something better - although a quick fix on the grid servers would probably be worthwhile in the meantime.
[12:08] Neas Bade: I think that if the grid services are discussed a bit upfront on the list, we can get a good push with a bunch of people doing code, and the break window will be smaller
[12:08] Adam Frisby: Yeah
[12:08] Adam Frisby: Woot
[12:08] Neas Bade: ah, adam managed to get cafinated
[12:08] Adam Frisby: Haha yes.
[12:08] Neas Bade: I saw that some earlier today
[12:09] Neas Bade: Tarevus, any ideas?
[12:09] Charles Krinkeb: There are UGA testing issues and Sim testing issues. It would be nice to seperate them. My thought is making less frequent UGA updates and allowing the Sim updates to occur at the sim owners discretion.
[12:09] Teravus Ousley: not on the grid services.. it hadn't been my focus .. so I have not studied the code much.
[12:10] Adam Frisby: Haha
[12:10] Finrod Meriman: i really hope my boss doesn't walk in and see this? :-)
[12:10] Neas Bade: I think that is going to be difficult, as client <-> grid issues may need to be synced
[12:10] Adam Frisby: *pulls out the poseballs*
[12:10] Neas Bade: oh, that brings up one last thing
[12:10] Neas Bade: it would be nice if a grid could mandate a minimum build version
[12:10] Neas Bade: in the client to grid handshake
[12:10] Adam Frisby: Yes, I'd like that too.
[12:10] Optik Philtr: oh yea
[12:11] Adam Frisby: Can we use the $revision$ tag somehow?
[12:11] Neas Bade: so you can force upgrade folks
[12:11] Charles Krinkeb: I would also. Perhaps in a config.xml setting?
[12:11] Adam Frisby: Is there anyway to get that updated on checkout?
[12:11] Adam Frisby: since $revision$ only updates on last file edit
[12:11] Neas Bade: Adam there should be
[12:11] Optik Philtr cheers
[12:11] More Wright: yeah I think in general we have few coders who really want to work on the grid servers or the networking side of things, or the udp networking code. I've been trying to find someone who wants to work on udp code for months but no luck
[12:11] Neas Bade: oh, right
[12:11] Adam Frisby: Grid servers I'll be interested in once I'm done with the stuff on my list
[12:11] Optik Philtr: lol
[12:11] Charles Krinkeb: on that note, getting the svn rXXXX to be knowable from the sim console or a header file would help. Preferably the former.
[12:12] Neas Bade: Adam, I'll research a way that we can get that number into the source
[12:12] Adam Frisby: Alright
[12:12] Charles Krinkeb: Right now, I write it down in my lab notebook, but others ask me this question from time to time
[12:12] Neas Bade: MW: I'm interested in working on the network side. It's been a while since I wrote UDP code, so it may take some ramp up
[12:13] Charles Krinkeb: I like this version of ODE, I hope it survives 24 hours.
[12:13] Teravus Ousley: :D
[12:13] Neas Bade: krinke, you should update when it crashes, teravus has more fixes in place
[12:13] Neas Bade needs to run off.
[12:13] Neas Bade: catch everyone on irc
[12:13] Charles Krinkeb: got it. With that said, maybe it wont crash
[12:13] Optik Philtr: ok
[12:14] Teravus Ousley: :D
[12:14] Teravus Ousley: Take care neas
[12:14] More Wright: Neas, cool, believe me it doesn't matter how rusty you are or how little someone knows about UDP networking, anything you do will be a great improvement on the current unloved code
[12:14] Charles Krinkeb: Anyone that wants to just chat, we can continue for a while, or head back to irc.
[12:15] Charles Krinkeb: Or, someone could, like, "build more stuff". We need "more stuff" built on this region.
[12:15] You: lol that picture cracks me up
[12:15] Charles Krinkeb: How did you do that Adam?
[12:15] Optik Philtr: ok one Starrax car comming up
[12:16] Teravus Ousley: hehe
[12:16] Optik Philtr: oh noes! I cant build! :(
[12:16] Adam Frisby shouts: Just me or is physics data not updating rotation?
[12:16] Teravus Ousley: it is.. but the're treated like cubes
[12:16] Optik Philtr: its rotating around Z axis
[12:16] Adam Frisby: Heh
[12:16] Charles Krinkeb: holy guacamole, batman. If that giant 8-ball rolls off the edge of the region, will we crash?
[12:16] Optik Philtr: lets find out...
[12:16] Adam Frisby: We need to build a cage!
[12:17] Optik Philtr: NOT!
[12:17] Teravus Ousley: .. also.. rotational velocity isn't being transmitted.
[12:17] Teravus Ousley: .. so the rotations 'step'
[12:17] Teravus Ousley: instead of interpolate
[12:17] Charles Krinkeb: Do we need a wall, or perhaps "billiard cushions" around the entire region?
[12:17] Adam Frisby: haha
[12:17] Teravus Ousley: hehe
[12:18] Teravus Ousley: well.. I'd suggest turning the prim off physical that hit the border.. because it's sending out terse updates for them even though they are disabled at -10m in the physics engine
[12:18] Charles Krinkeb: I would like to leave the balls rolling around, but perhaps someone can build a 2 meter high wall around the region this afternoon?
[12:18] More Wright: some one recurse adam
[12:18] You: ok
[12:18] More Wright: ahh hes back
[12:18] Adam Frisby: Haha
[12:19] More Wright: rescue even
[12:19] Charles Krinkeb: Hey, Adam. This is a long ways from "Mortville"??
[12:19] Optik Philtr: yay physics FTW!
[12:19] Adam Frisby: Yeah
[12:19] Adam Frisby: Certainly getting better every time I viist
[12:19] Charles Krinkeb: I think back to last May and say "This is *real* progress"
[12:20] Optik Philtr: so im confused is permision switch enabled in wright?
[12:20] Adam Frisby: hgahga
[12:20] Charles Krinkeb: Only "Plaza Builder" can build here.
[12:20] Adam Frisby: Uh oh
[12:20] You: lol
[12:20] Adam Frisby: You have permissions on?
[12:20] Charles Krinkeb: You can morph any prim you like.
[12:21] Optik Philtr: oh and copy
[12:21] You: ROFL
[12:21] Optik Philtr: that will make the car fun
[12:21] Adam Frisby shouts: Just me or is the ground really rubbery?
[12:21] You: its me
[12:21] Charles Krinkeb: Copy only. This was done for reliability. I or nebadon can give the "Plaza Builder" password to anyone that is reliable.
[12:21] You: moving the prim
[12:21] You: lol
[12:22] Adam Frisby shouts: Nah not that
[12:22] Adam Frisby shouts: watch
[12:22] Charles Krinkeb: Are any of you guys "reliable"??
[12:22] Adam Frisby: Hrrm, stopped doing it.
[12:22] Adam Frisby: No!
[12:22] Charles Krinkeb: Alright, give it to Adam first, then
[12:22] Adam Frisby: I think I killed the tennis ball
[12:22] Adam Frisby: Oh there it is
[12:23] Teravus Ousley: move it before setting it physical
[12:23] Charles Krinkeb: In any case, this mostly affects terraforming. You can copy any prim you wish and change it to any texture you wish. We just dont want the land to change as that makes it more unreliable.
[12:23] Teravus Ousley: heh.. otherwise it'll collide with it's ghost
[12:23] Adam Frisby: OK, I'm going back to drudge-work. :)
[12:23] You: the sim held up great
[12:23] Teravus Ousley: (fixed in current revision)
[12:23] Adam Frisby: Then I might build myself something on OpenSim. :)
[12:23] You: with all the ODE stuff happening
[12:24] Teravus Ousley: DON'T resize it!
[12:24] Adam Frisby: Btw - if anyone feels like adding physics shells for sculpties - there's US$200 in it from me.
[12:24] Teravus Ousley: Optic.. don't resize that prim :P
[12:24] Teravus Ousley: .. it'll crash the sim on this revision
[12:24] danx2 danx0r: "feel so lonely..."
[12:25] More Wright: still no one done your sculpties physics ?
[12:25] Adam Frisby: Nope
[12:25] Adam Frisby: and I've got big floating islands I want to build.
[12:25] More Wright: should only take a few hours, thought someone would do those
[12:25] Adam Frisby: Yeah, me too. heh.
[12:25] Adam Frisby: Was going to think about doing it myself. hehe.
[12:26] Charles Krinkeb: Ter: After we all go, would you like me to update this region to the latest?
[12:26] More Wright: yeah all the details someone would need is on the SL wiki, certainly easier to do them than cuts in prims etc
[12:26] You: that would be good i think
[12:26] You: it would prevent some crashing
[12:27] Charles Krinkeb: I'll just wait until its quiet a little later this afternoon.
[12:27] You: right now if you resize a cut out cylinder
[12:27] Teravus Ousley: makes sense. Or if it crashes.
[12:27] You: it will kill the sim
[12:27] Charles Krinkeb: Whichever occurs first, I think I have a round "Tuit" in my desk drawer.
[12:27] Teravus Ousley: hehe.. I had one on my wall
[12:28] Charles Krinkeb: I just dont want to stop the fun. This is my reward, watching all us children playing for once.
[12:28] Adam Frisby: Hehe
[12:28] Adam Frisby: OK, I'm heading off.
[12:28] Adam Frisby: Seeya fokls
[12:28] You: later adam
[12:28] Teravus Ousley: Take care Adam :D
[12:28] Charles Krinkeb: Bye adam
[12:29] Teravus Ousley: I'm as excited as you are cfk
[12:29] Charles Krinkeb: I feel just like a little kid with my first toy all of a sudden
[12:29] More Wright: we should be able to grab another person/avatar and throw them around
[12:29] Optik Philtr: hehee
[12:29] Charles Krinkeb: Throw me around
[12:29] Teravus Ousley: client limited in that aspect.. client won't generate grab events on avatar
[12:29] More Wright: I think danx0r's avatar would bounce around nicely
[12:30] Teravus Ousley: Yes, that's another thing I've got to work on.. figuring out how to change an avatars physics height by it's actual height.. I don't quite know about how to get that info now though :)
[12:31] You: lol nice
[12:31] Charles Krinkeb: I would suggest we make new building outside the current walkways as I would like to appreciate nebadon's beautiful corporate plaza for a long time.
[12:31] Optik Philtr: yay our first griefer weapon
[12:31] You: ROFL
[12:31] Optik Philtr: a cube lol
[12:31] Charles Krinkeb: And some day, I hope to see the water flowing in the fountain.
[12:31] You: Lag Cube
[12:31] Teravus Ousley: hehe.. cubecage.. as it will be immortally known from this point on
[12:32] More Wright: yeah we just need to get clouds working then we can have rain and turn the fountains back on
[12:32] You: lol
[12:32] Optik Philtr: yes this mandantory no water use is getting old
[12:32] You: yea im starting to see brown patches in the grass
[12:32] Teravus Ousley: just like Adam is the first person to do a head stand in OpenSim
[12:33] You: damn that gives me an idea for SL
[12:33] You: automated sprinkler system
[12:33] Optik Philtr: lol camping gym equipment
[12:33] Charles Krinkeb: Can you guys help me think of and build out the "Sim Owners" garden on the other side of the building sometime?
[12:33] Optik Philtr: oh cool
[12:33] You: yea we need a nice sculpty tree
[12:33] You: i might work on that later
[12:34] Optik Philtr: well my sculptypaint is getting an error
[12:34] Charles Krinkeb: Also, we need a grotto, or some other type of garden of a little mini-forest on the other side, maybe towards the edge?
[12:34] Optik Philtr: i have some crude ones ill give ya for peices
[12:34] Optik Philtr: but i dont know how to get them to you
[12:35] Charles Krinkeb: I am hoping this region will continue to evolve as our meeting place until it is wall-to-wall with prims and avatars. Then, we find a quieter place, but not until then.
[12:35] Charles Krinkeb: Just build stuff here
[12:35] Optik Philtr: yea the more it looks like a linden sandbox the better
[12:35] Optik Philtr remakes the blocky green tank
[12:36] Charles Krinkeb: Building stuff consists of merely copying a prim and morphing it.
[12:36] More Wright: we need to do another stress test sometime
[12:36] Charles Krinkeb: ld wants to do one on Saturday on Apex.
[12:36] Teravus Ousley: lol, yes.. I've done about 95 physical prim. It does consume 100% of the CPU and the physics slows down. a tad.
[12:36] More Wright: sounds ok, I'm away for a couple of days, but back by saturday
[12:36] Charles Krinkeb: We have the record at 20 avatars currently. We also know that 10 avatars have been in each of our office hours.
[12:37] More Wright: yeah a couple of weeks ago we had 13 in office hours
[12:37] Charles Krinkeb: Maybe we need to define a second time during the week, perhaps call it "play hours" or "Testing Hours" and encourage all to come in here at the same time.
[12:38] Teravus Ousley: physical prim are interesting.. because they take a lot less CPU time once they've settled
[12:38] Charles Krinkeb: It might be a different group then the developers, but that's Ok. I think we should consider defining a "Testing Hours".
[12:38] More Wright: yeah sounds ok
[12:38] More Wright: /good
[12:39] Charles Krinkeb: 11AM PST Saturday's, maybe, which is, I think, 1700 UTC??
[12:39] Charles Krinkeb: Very good, nebadon
[12:40] Charles Krinkeb: I dont want to steal ld's thunder, so I will not start it this weekend.
[12:42] You: lol
[12:42] Charles Krinkeb: I melted into the ground just standing there.
[12:43] Teravus Ousley gets off that cube
[12:43] Charles Krinkeb: Is this a chair?
[12:43] You: yea
[12:43] You: sculpty
[12:44] You: yea sit dont work to well yet
[12:44] You: hehe
[12:44] More Wright: we need to make the hippo physical and drop it on people
[12:44] Charles Krinkeb: Thats Ok. Its just the first chair I have seen in OpenSim
[12:44] You: lol
[12:44] Teravus Ousley: I've got a few things to do guys.. so I'll be on IRC ;D Thanks for a lot of fun
[12:45] You: alrighty
[12:45] You: i'll post the meeting notes on the Wiki